1. Field of the Invention
The present invention relates to a technique for inspecting printed documents. More specifically, the present invention relates to a technique for detecting a print defect of an image read from a printed document.
2. Description of the Related Art
There is a technique for inspecting a printed document to detect a print defect of the printed document. In this technique, based on an original image input into a printing apparatus, printing processing is performed. A reading apparatus scans and reads the printed document output from the printing apparatus to obtain an image thereof. Then, from the read image and the input image, feature points thereof are extracted. Using the feature points, position adjustment thereof is performed. Then, comparison processing of the read image and the input image after the position adjustment is performed. Thus, a printing defect is detected.
Japanese Patent Application Laid-open No. 2010-249565 discusses a position adjustment technique in which two steps of position adjustment processing is performed on images each divided into a plurality of areas. The technique discussed in Japanese Patent Application Laid-open No. 2010-249565 performs a first step position adjustment of specific areas to perform rough position adjustment thereof. Consequently a second position adjustment is performed to perform accurate position adjustment of the specific areas.
It is necessary for performing position adjustment processing to extract feature points for position adjustment. Therefore, if the information of the extracted feature points (e.g. the number of the extracted feature points and/or the position information of the extracted feature points) is not appropriate or sufficient to perform position adjustment processing, accuracy of the position adjustment is deteriorated.
For example, in order to reduce the capacity of the buffer memory for a read image of a printed document or to perform a real time inspection, in a case where images sequentially read by a reading apparatus is inspected sequentially for a band region of a predetermined size, sometimes information of feature points within a band region is not enough depending on the band region. That is because an image within a band region may be configured of faint characters or gradation. Therefore, even if two steps of position adjustment processing are performed on such a band region, the accuracy of the position adjustment is not improved. Thus, position adjustment processing with enough accuracy may not sometimes be performed.